## Fluids

• rsaurabh
Subscriber

Hi all,

i did a pulsatile flow study over 3 different grids and simulated my case in Fluent. Now i have all results for all 3 grids that i want to check and analyze but i am facing an issue of how will i report % error in grids.

for showing grid independency, i have plotted velocity profile at few sections and it is clearly visible. but i need to report errors included in simulation and i do not have any experimental data to validate my results. how can i deal with it?

any suggestions are welcome.

Regards

Saurabh

• peteroznewman
Subscriber

I read a paper "Use of the FDA nozzle model to illustrate validation techniques in CFD simulations" that I used to answer your question. One of the authors works at ANSYS.

A Standard for calculating error in CFD models has been created. (I have read the Standard for Computational Solid Mechanics).
[14] ASME V&V 20-2009 Standard for Verification and Validation in Computational Modeling of Fluid Dynamics and Heat Transfer.

If you have no experimental data, there are two uncertainties you can study: numerical uncertainty and input parameter uncertainty.

Numerical uncertainty (Unum). The Grid Convergence Index (GCI) method, based on Richardson extrapolation theory, was adopted for conducting the grid convergence study and for obtaining the numerical uncertainty due to discretization of the flow domain into finite volumes.

The grid size h is estimated from (Vol/N)^1/3 where N is the total number of finite volume elements and Vol is the total volume of the flow domain.

The grid refinement factor is defined as w21 = h2/h1, where h2 and h1 are grid sizes that result in medium-density and high-density grids respectively.

The low-density grid size is h3 and the change in size should be a constant factor, w = w21 = w32 where the factor w should be between 1.3 and 2.

I found a paper on the internet on Grid Convergence Study, but of course it uses different nomenclature.

Input parameter uncertainty analysis.

The uncertainty in the quantities of interest (i.e. simulation outputs) due to uncertainty in the simulation input parameters can be estimated using the sensitivity coefficient method for uncertainty propagation [14]. This method is a local approach which assesses the effect of a small perturbation in the input parameter(s) on each simulation output parameter. The ASME V&V20-2009 recommends using Monte Carlo analysis to provide a global estimate of the input parameter uncertainty [14].